Log in

No account? Create an account
27 November 2016 @ 01:59 pm
Filter Bubbles and Echo Chambers...  
This started as a Facebook link post, but as it went on I figured it made more sense as a blog post. I am reacting to this opinion column from back in October. Although some of these thoughts have been stewing longer. I meant to post this some time ago about the idea that filter bubbles and echo chambers of search engines and social media may be overstated that we mostly read the same news story etc.

I can well believe that the most powerful filter on our perceptions is the one between our ears, so that we see or hear the same things and take away radically different things. I experience this occassionally.

On the other hand I can see the credibility of the filter bubble idea and echo chamber ideas since what we need and what we want are so often divergent, it is easy to believe that a search engine or social media site that gives us what we want is therefore diverging from giving us what we need. An opinion column like this can no doubt cherry pick its studies to make its point etc. so does not negate that the prima facia case for filter bubbles.

Yet the history of search engines has intrigued me (I read this book). Back in the day we had search engines like Alta Vista that could find instances of whatever you searched for, but not particularly salient ones. The success of Google was that people noticed how much better its search results were and switched over to it from search engines that were previously far more dominant (so search engine usage is not explicable by pure inertia). I remember I actually held on not using Google longer than most (I have a stubborn streak).

Filter bubbles as popularly explained seem to imply sub-optimal search results (relevant pages that would inform us about something etc. fail to show up because of distaste or whatever) and that should create an opening for a better search engine to come along and challenge Google or whoever's dominance. This could be explained by positing that Google (or Bing etc.) is near a fitness peak and so even if it is less than optimal it is close enough that the degraded efficacy does not create pressure to shift away, this still limits how much filter bubbles could effect search. Or we could posit that while past search engine choice was driven by performance, now inertia rules because the search market has matured or whatever giving filter bubble free rein. Or that while on one metric Google Search is sub-optimal on other metrics it is so far ahead of the pack that it need not fear being disabled by filter bubbling. What exactly is going on is hard to say without a lot more investigation.

So filter bubbles look like they could be a thing on first pass, but a closer examination reveals there are factors that make it a more complicated proposition and so on. Hence why I find this sort of article interesting.

I have far fewer thoughts about social media echo chambers, I admit that my Facebook is a rather specific subsection of humanity politically (and culturally), but judging from comments on my friends posting there is a fair bit of range for many of them.

In case you get cut out by Globe and Mail's paywall here are the sources the above article sites:
Andrew M. Guess of NYU.
Will the Global Village Fracture Into Tribes? Recommender Systems and Their Effects on Consumer Fragmentation Hosanagar, Fleder, Lee and Buja.

EDIT: December 6, 2016, I should add that one particular thing that has led me to some of the above thoughts on filter bubbles is this paper by my friends Isaac Record and Boaz Miller. I notice as I look for the link that another friend of mine Mike Thicke had made some blog posts about this that I read and that may have been swirling in my head as I composed this.

Really we need to do an experiment where we get people to use a set of nerfed Search Engines (with a secret non-nerfed Engine) for a couple of weeks and see whether they can figure out which ones we nerfed just by the results they get in searches they were doing anyway as part of life. I suspect people would be highly sensitive to sub-optimal search, which might indicate a limit on the effect of filter bubbles.
Current Location: Oakville, Ontario, Canada
Current Mood: quizical

Recent Posts from This Journal